1.1 What is Statistics?

(Page 1 of 15)

1.1 What is Statistics?

Definition of Statistics
Statisticsisthestudyofhowtocollect,organize,analyze,and
interpretnumericalinformationfromdata.Descriptivestatistics
involvesmethodsoforganizing,picturingandsummarizing
informationfromdata.Inferentialstatisticsinvolvesmethodsof
usinginformationfromasampletodrawconclusionsaboutthe
population.
KeepinMind:
1*
2*

Statisticalinferencesarenomoreaccuratethanthedatathey
arebasedon(weakestlink).
Statisticalresultsshouldbeinterpretedbyonewho
understandsthemethodsusedaswellasthesubjectmatter.

Individuals and Variables

Individualsarethepeopleorobjectsincludedinthestudy.A
variableisthecharacteristicoftheindividualtobemeasured
orobserved.
Forexample,ifwewanttodoastudyaboutthepeoplewhohave
climbedMt.Everest,thentheindividualsinthestudyarethe
actualpeoplewhomadeittothetop.Thevariablestomeasure
orobservemightbetheheight,weight,race,gender,income,etc
oftheindividualsthatmadeittothetopofMt.Everest.
Variables: Quantitative vs. Qualitative
Aquantitativevariablehasavalueornumericalmeasurementfor
whichoperationssuchasadditionoraveragingmakesense.A

qualitativevariabledescribesandindividualbyplacingthe
individualintoacategoryorgroupsuchasmaleorfemale.

1.1 What is Statistics? (Page 2 of 15)

Example A
Statewhetherthedataisqualitativeorquantitative.
1.
Thecolorofapersonseye.
2.

Theheightofapersonininches.

3.

Thea/b/c/dresponsesonaquestionnaire.

Population Data and Sample Data

Inpopulationdata,thevariableisfromeveryindividualof
interest.Insampledatathevariableisonlyfromsomeofthe
individualsofinterest.
Guided Exercise 1
TelevisionstationQUEwantstoknowtheproportionofTV
ownersinVirginiawhowatchthestationsnewprogramatleast
onceaweek.Thestationaskedagroupof1000TVownersin
Virginiaiftheywatchtheprogramatleastonceaweek.
1.
Identifytheindividualsinthestudy.
2.

Identifythevariable.

3.

Dothedatacompriseasample?Ifso,whatistheunderlying
population?
Yes.Theimpliedpopulationistheresponses(watch/not
watch)ofallTVownersinVirginia.

4.

Isthevariablequantitativeorqualitative?

5.

Identifyaquantitativevariablethatmightofinterest.

1.1 What is Statistics? (Page 3 of 15)

Example 1
SupposetheHawaiiDepartmentofAgriculturewishesto
conductastudyofpineapplesinanexperimentalfield.
1.
Whataretheindividualsinthestudy?
2.

Whataresomepossiblequantitativevariablesofinterest?
Theweightofeachpineappleintheexperimentalfield.
Themaximumgirthofeachpineappleinthe
experimentalfield.

3.

Whataresomepossiblequalitativevariablesofinterest?
The taste [color] of every pineapple in the field.

1.1 What is Statistics? (Page 4 of 15)

Levels of Measurement

2.

NominalLevel(innameonly):Qualitieswithno
ranking/ordering;nonumericalorquantitativevalue.Data
consistsofnames,labelsandcategories.
Taos, Acoma, Zuni and Cochiti are names of four
nativeAmericanpueblos.
Carcolorsforacertainmodelare:red,silver,blueandblack.

2.

OrdinalLevel:Canbearrangedinsomeorder,butthe

1.

1.

1.
2.

differencesbetweenthedatavaluesaremeaningless.
Of17fishingreelsrated:6wereratedgoodquality,4were
ratedbetterquality,and7wereratedbestquality.
Outofahighschoolclassof319,Walterranked4th,June
ranked12th,andJimranked20th.

3.

1.
2.
3.

IntervalLevel:Datavaluescanberankedandthedifferences
betweendatavaluesaremeaningful.However,thereisno
intrinsiczero,orstartingpoint,andtheratioofdatavaluesare
meaningless.Note:CalendardatesandCelsius&Fahrenheit
temperaturereadingshavenomeaningfulzeroandratiosare
meaningless.
Theyearsinwhichdemocratswonpresidentialelections.
Body temperature in degrees Celsius (or Fahrenheit) of
troutswimmingintheNorthRiver.
BuildingAwasbuiltin1284,BuildingBin1492andBuilding
Cin5bce.

4.

1.
2.
3.

RatioLevel:Similartointerval,exceptthereisatruezero,or
startingpoint,andtheratiosofdatavalueshavemeaning.
CoretemperatureofstarsmeasuredindegreesKelvin.
Timeelapsedbetweenthedepositofacheckandtheclearance
ofthatcheck.
LengthoftroutintheNorthRiver.

1.1 What is Statistics? (Page 5 of 15)

Levels of Measurement
NominalLevel(innameonly):Qualitieswithno
ranking/ordering;nonumericalorquantitativevalue.
2.
OrdinalLevel:Canbearrangedinsomenumericalorder,but
thedifferencesbetweenthedatavaluesaremeaningless.
3.
IntervalLevel:Datavaluescanberankedandthedifferences
betweendatavaluesaremeaningful.However,thereisno
intrinsiczero,orstartingpoint,andtheratioofdatavaluesare
meaningless.
4.
RatioLevel:Similartointerval,exceptthereisaninherent
zero,orstartingpoint,andtheratiosofdatavalueshave
meaning.
Guided Exercise 2
1.

Statethelevelofmeasurementforeachofthefollowing:
1.
ThesenatorsnameisSamWilson.
2.

Thesenatoris58yearsold.

3.

Thesenatorwaselectedin1963,1969,1981,and1994.

4.

Histaxableincomeis$278,314.19

5.

Of1100votersinhisdistrict:400stronglyfavorhisbill;300
favor;200neutral;150donotfavor,and50stronglydonot
favorhisbill.

6.

Thesenatorismarried.

7.

Thesenatorhaddivorcesin1965and1982.

8.

Anewspaperrankedthesenator7thforhisvotingrecordon
publiceducation.

1.2 Random Samples (Page 6 of 15)

1.2

Random Samples

Simple Random Sample

Asimplerandomsampleofnmeasurementsfromapopulation
isoneselectedinsuchamannerthat
1.
every sample of size n from the population has
equalprobabilityofbeingselected,and
2.
every member of the population has equal probability of
beingincludedinthesample.
Example C
ConsiderthepopulationofallcoyotesinthewesternU.S.The
sampleofthatpopulationthatranchersobserveislargelythe
coyotesthatprefertoliveneararanch;theyalsoliketoeatlamb.
Theranchersconcludedthatallcoyotesaredangerousandgotthe
governmenttoassistindistributingapoisonbaittoreducethe
overallcoyotepopulation.Theoverallpopulationofcoyoteswas
reduced,buttheranchersstilllostalmostasmanysheepas
before.Why?
1.
Isthesampletheranchersobservedarandomsample?
2.

If not, is it safe to use the results to describe the

entirepopulation?

3.

Is the idea of reducing the size of the entire

populationjustifiedbasedontheranchersexperience?

Note:Furtherstudyhasshownthatcoyoteswhoeatsheepare
consistentintheirpreferenceforsheep,whereasthe
majorityofcoyotesinthewildsticktofoodsfoundin
thewild.

1.2 Random Samples (Page 7 of 15)

Example D
Dothefollowingproceduresgivearandomsamplefortheentire
populationofNewYorkCity?Why/Whynot?
1.
Selecteverythirdwomanenteringabeautyshop.
2.

Selecteverythirdpersoncomingoutofaboxingmatchat
MadisonSquareGarden.

Example 6: Random Number Table

Todetermineifthelatestshipmentof500Toyotasmeetsemission
standards,arandomsampleof30ischosenandtested.Howcan
youmakesurethesampleof30chosenisarandomsample?See
therandomnumbertableinAppendixII,pageA9.
92630
79445
28703
17138

78240
78735
51709
14280

19267
71549
94456
51333

95457
44843
95761
51287

53497
26104
33393
99281

23894
67318
99411
81017

33708
00701
25270
55137

23501

85846

88472

64688

37818

96593 28661

Simulation
Asimulationisanumericalfacsimileorrepresentationofareal
worldphenomenon.
Random Number Generator on the TI-83
1.
2.

MATH/PRB/1:rand
MATH/PRB/5:randInt(min,max,numberofintegers)

1.2 Random Samples (Page 8 of 15)

Example E
Usearandomnumbertabletosimulateeachofthefollowing.
1.
Choosethenumbersforthenextlottery.Thatis,randomly
choosesixnumbersfrom1to52.
92630
79445
28703
17138

78240
78735
51709
14280

19267
71549
94456
51333

95457
44843
95761
51287

53497
26104
33393
99281

23
67
99
81

23501

85846

88472

64688

37818

96

b.Theoutcomesoftossingadie20times.

3.

92630
79445
28703
17138

78240
78735
51709
14280

19267
71549
94456
51333

95457
44843
95761
51287

53497
26104
33393
99281

23
67
99
81

23501

85846

88472

64688

37818

96

Assignthelettersa,b,c,d,oreasthecorrectresponseon
a10questionmultiplechoiceexam.
92630
79445
28703
17138

78240
78735
51709
14280

19267
71549
94456
51333

95457
44843
95761
51287

53497
26104
33393
99281

23
67
99
81

23501

85846

88472

64688

37818

96

1.2 Random Samples (Page 9 of 15)

Other Sampling Techniques

1.StratifiedSampling
Stratify(divide)thepopulationbyacommoncharacteristic(such
asage,class,gender,etc.)andtakearandomsampleofeach
stratum[ofteninaccordancetotheirpercentofthepopulation].
e.g.dividethepopulationbyraceandsurveyanumberof
randomlyselectedindividualsfromeachrace(oftenin
thesameproportiontheyoccurintheoverallpopulation).
2.SystematicSampling
Usedwhentheelementsofthepopulationarearrangedinanatural
sequentialorder.
e.g.takeevery5thpersoncomingthroughacafeterialine.
3.ClusterSampling
Usedextensivelybygovernmentandresearchorganizations.
Randomlyselectasampleofpreexistingsectionsor
clusters(oftengeographicsections).Everymemberofthe
clusterisincludedinthesample/survey.
e.g.randomlyselect30schoolsandsurveyeverystudent
ineachschool.
4.ConvenienceSampling
Usesdatathatareconvenientlyandreadilyobtained.
e.g.walkoutsideandsurveythefirst100peoplethatwill
talktoyou.

1.3 Experimental Design (Page 10 of 15)

1.3 Experimental Design

Basic guidelines for planning a statistical study
1.
2.
3.
4.
5.

6.

Identifytheindividualsorobjectsofinterest.
Specifythevariablestomeasureorobserve.
Determineifyouwillusethepopulationorarepresentative
sample.Decidesonaviablesamplingmethod.
Collectthedata.
Usetheappropriatedescriptivestatisticsmethods(Chapters
2, 3 and 10) and make decisions using the appropriate
inferentialstatisticsmethods(Chapters812).
Noteanyconcernyoumighthaveaboutyourdatacollection
methodsandlistanyrecommendationsforfuturestudies.

Ways to Produce Data

1.
2.

3.
2.
3.

Census: Measurements or observations of the entire

population.
Sampling:Measurementsorobservationsfroma
representativepartofthepopulationi.e.simple
randomsample.
Simulation:Numericalmodelingofrealworldphenomena.
Experiment:Imposeatreatmentandmeasure/observethe
changeinthevariableofinterest.
ObservationalStudy:Observationsandmeasurementsare
madeinawaythatdoesnotchangetheresponseorthe
variablebeingmeasured.

1.3 Experimental Design (Page 11 of 15)

Example F
Whichdatagatheringtechniquemightbebestsuitedforeachof
thefollowingsituations?
1.
Studytheeffectofstoppingthecoolingprocessina
nuclearreactor.
2.

Studytheamountoftimecollegestudentstaking
afullcourseloadwouldspendwatchingTV.

3.

Studytheeffectofacalciumsupplementgiven
toyounggirlsonbonemass.

4.

Studythecredithourloadofeachstudentenrolled
atPalomarattheendoftheadd/dropperiod.

1.3 Experimental Design (Page 12 of 15)

Experiment and Observational Study

Inanobservationalstudy,observationsandmeasurementsof
individualsareconductedinawaythatdoesnotchangethe
responseorthevariablebeingmeasured.Inanexperiment,a
treatmentisdeliberatelyimposedontheindividualsinorderto
observeapossiblechangeintheresponseorvariablebeing
measured.Inastudy,oftentheindividualsaredividedintotwo
groupsacontrolgroupandanexperimentalgroup.The
experimentalgroupisactuallygiventhetreatmentandthe
controlgroupisnotgiventhetreatment.
Example 8
In1778CaptainJamesCookintroducedgoatstotheHawaiian
Islands.ItwaslaterobservedthattheSilverSwordplantappeared
tobelessandlesscommon.Botanistssuspectedthegoatstobethe
causeandconductedastatisticalstudy.Theysetupstations
aroundtheislandswithsimilarclimateandsoilconditions.Each
stationconsistedoftwoplotsofland,onewithafencearounditto
keepthegoatsout.Identifythe
1.
treatment
2.

experimentalgroup

3.

controlgroup

1.3 Experimental Design (Page 13 of 15)

Randomized 2-Treatment Experiments Placebo Effect

Theplaceboeffectoccurswhenasubjectinthecontrol
groupreceivesnotreatment,butbelievessheisinfact
receivingtreatmentandrespondsfavorably.
Example 9
Formorethanadecadedoctorshavebeenusinglaserstodrill
holesinthehearttoreduceangina(chestpain).Manypatients
reportedlastingandsignificantrelieffromtheprocedure.Totestif
thereliefisduetotheplaceboeffectornotarandomizedtwo
treatmentexperimentwascompleted.Agroupof298volunteers
withsevere,untreatablechestpainwererandomlyassignedtoget
thelasertreatmentornot.Thepatientsweresedatedbutawake.
Theycouldhearthedoctorsdiscussthelaserprocess.Eachpatient
thoughtheorshewasreceivingthetreatment.
Conclusion:Thelaserpatientsdidwell.Butshockingly,the
placebopatientsshowedmoreimprovementinpainrelief.
Themedicalimpactsofthisstudyarestillbeinginvestigated.
Thestudyofexample9hasmanyfeaturesofgoodexperimental
design.Thereisacontrolgroupwhoreceivedthedummy
treatmentandtheexperimentalgroupwhoreceivedtheactual
treatment.Thecontrolgroupisusedtoaccountforinfluenceof
lurkingorconfoundingvariablesthatmightaccountforsome
ofthechangesobserved.
Randomizationisusedtoassignindividualstothetwotreatment
groups.Thishelpspreventbiasinselectingmembersforeach
group.
Replicationoftheexperimentonmanypatientsreducesthe
possibilitythatthedifferencesinpainreliefforthetwogroups

occurredbychancealone.

1.3 Experimental Design (Page 14 of 15)

Double-Blind Experiment
Adoubleblindexperimentisoneinwhichneitherthepatients,
northeobserversknowwhichsubjectsarereceivingthe
treatment.Theyhelpcontrolthebiasesofdoctorsandresearchers.
Survey Questions and their Pitfalls
Gatheringdatabysimplyaskingpeoplequestionsistheessenceof
asurvey.
Cautions with Survey Questions
1.
2.
3.
4.
5.
6.
7.

8.

9.
10.

Canyouconvertthedatatonumbers?
Isthewordingofthequestionunbiased?
Voluntary responses often overrepresent strong
[negative]opinions.
Canyouexpectatruthfulresponse?
Isthesamplerepresentativeofthepopulation?
HiddenBias:Howwouldyoudesign/administerasurveyof
PalomarCollegeStudents?
OtherVariablesmayestablishaCauseEffectRelationship.
Since,ingeneral,eventswithhigherticketpriceshavehigher
attendance,shouldyouraiseticketpricestoincreaserevenue
OverGeneralizingResults
Results of drug experiments on lab rats cannot be
generalizedtootheranimals.
Howdoyouhandleasignificantproportionofnonresponses.
Pollregisteredversuslikelyvoters.

1.3 Experimental Design (Page 15 of 15)

Guided Exercise 4
Commentontheusefulnessofdatacollectedasdescribed:
1.

A uniformed law officer asks a group of college freshman

theirnameandifhe/shehasuseddrugsinthelastmonth.

2.

Janesawdatathatshowthatcitieswithmorehomeless
peoplehavemorelowincomehousing.Doesbuildinglow
incomehousingproducemorehomelessness?

3.

Asurveyaboutfoodinthecafeteriawasconductedby
havingsurveyformsavailableattheregister.Adropboxfor
theformsisoutsidethecafeteria.

4.

Extensive studies on coronary problems were conducted

usingmenoverage50asthesubjects.
PDF to Word